Energy storage power station performance estimation method and system

Abstract

The application discloses a kind of energy storage power station performance estimation methods, including obtaining the historical data information of target energy storage power station;Energy storage state variable, dispatching action variable and grid environment variable information of target energy storage power station are constructed to construct training data set;Based on state space, state coupling and neural operator scheme, target energy storage power station performance estimation initial model is constructed and target energy storage power station performance estimation model is trained;Using the obtained model, the performance estimation of target energy storage power station is completed.The application also discloses a kind of system for implementing the energy storage power station performance estimation method.The application not only can realize the performance estimation of energy storage power station, but also has higher reliability, better accuracy.

Description

Technical Field

[0001] This invention belongs to the field of electrical automation, and specifically relates to a method and system for estimating the performance of an energy storage power station. Background Technology

[0002] With economic and technological development and the improvement of people's living standards, electricity has become an indispensable secondary energy source in people's production and daily life, bringing endless convenience. Therefore, ensuring a stable and reliable supply of electricity has become one of the most important tasks of the power system.

[0003] Energy storage power stations typically play a role in energy regulation and power support within power systems. They can charge, discharge, or regulate reactive power according to grid dispatch needs, and are commonly used in scenarios such as peak shaving and valley filling, renewable energy integration, frequency regulation, voltage support, and reserve support. Currently, more and more renewable energy generation systems are being integrated into the power system, making energy storage power stations increasingly important. Therefore, estimating the performance of energy storage power stations is becoming increasingly crucial.

[0004] Currently, commonly used performance estimation methods for energy storage power stations include fixed-index weighted estimation methods, rule-based estimation methods, and machine learning-based estimation methods. Fixed-index weighted estimation methods typically estimate the performance of energy storage power stations by weighting and summing indicators such as SOC, SOH, output power, tracking error, temperature, and available capacity. While simple to implement, these methods struggle to reflect changes in the meaning and weight of indicators under different operating conditions, resulting in poor accuracy. Rule-based estimation methods rely on manually set thresholds and empirical rules, offering some interpretability under simple operating conditions, but lack adaptability in multi-task switching scenarios such as frequency regulation, peak shaving, standby, and voltage support. Machine learning-based estimation methods can extract complex features, but existing solutions often employ general neural networks, attention mechanisms, or gating mechanisms for feature fusion, easily overlooking the physical consistency between power response, SOC evolution, temperature accumulation, and health degradation within the energy storage power station. Summary of the Invention

[0005] One of the objectives of this invention is to provide a highly reliable and accurate method for estimating the performance of energy storage power stations.

[0006] The second objective of this invention is to provide a system for implementing the aforementioned energy storage power station performance estimation method.

[0007] The energy storage power station performance estimation method provided by this invention includes the following steps:

[0008] S1. Obtain historical data information of the target energy storage power station;

[0009] S2. Based on the data obtained in step S1, construct the energy storage state variables, scheduling action variables, and power grid environment variables of the target energy storage power station to build a training dataset;

[0010] S3. Based on the state-space, state-coupled, and neural operator schemes, construct an initial model for performance estimation of the target energy storage power station;

[0011] S4. Using the training dataset obtained in step S2, train the model constructed in step S3 to obtain the performance estimation model of the target energy storage power station;

[0012] S5. Using the model obtained in step S4, complete the performance estimation of the target energy storage power station.

[0013] Step S1, which involves obtaining historical data information of the target energy storage power station, specifically includes the following steps:

[0014] Obtain historical data information of the target energy storage power station;

[0015] The data information includes the active power data, reactive power data, state of charge data, health status data, battery temperature data, current data, and terminal voltage data of the target energy storage power station; active power command data, reactive power command data, ramp speed requirement data, and operating mode data received by the target energy storage power station from the connected power grid; and frequency data, grid connection point voltage data, frequency deviation data, voltage deviation data, load power data, renewable energy output data, and ambient temperature data of the power grid to which the target energy storage power station is connected.

[0016] Step S2, which involves constructing the energy storage state variables, scheduling action variables, and grid environment variables of the target energy storage power station based on the data information obtained in step S1, to build a training dataset, specifically includes the following steps:

[0017] Construct the energy storage state variables of the target energy storage power station at time t :

[0018] In the formula Let be the active power of the target energy storage power station at time t; Let be the reactive power of the target energy storage power station at time t; The state of charge of the target energy storage power station at time t; The health status of the target energy storage power station at time t; Let be the battery temperature of the target energy storage power station at time t; Let be the equivalent internal resistance of the target energy storage power station at time t; Let be the current of the target energy storage power station at time t; Let be the terminal voltage of the target energy storage power station at time t;

[0019] Construct the scheduling action variables of the target energy storage power station at time t. :

[0020] In the formula The active power command received by the target energy storage power station at time t; The reactive power command received by the target energy storage power station at time t; The required ramp rate for the target energy storage power station at time t; The variable is the operating mode identifier of the target energy storage power station at time t. The operating modes include frequency regulation mode, peak shaving mode, standby mode, voltage support mode, standby mode, and power limiting operation mode.

[0021] Construct the grid environment variables of the target energy storage power station at time t :

[0022] In the formula The frequency of the power grid to which the target energy storage power station is connected at time t; The grid connection point voltage of the power grid to which the target energy storage power station is connected at time t; The frequency deviation of the power grid to which the target energy storage power station is connected at time t; The voltage deviation of the power grid to which the target energy storage power station is connected at time t; The load power of the power grid connected to the target energy storage power station at time t; The renewable energy output value of the power grid connected to the target energy storage power station at time t; The ambient temperature of the power grid to which the target energy storage power station is connected at time t;

[0023] Construct the evaluation window at time t for ,in The length of the evaluation window;

[0024] Construct the comprehensive input sequence of the target energy storage power station at time t for ,in Indicated in the evaluation window All energy storage state variables within, Indicated in the evaluation window All scheduling action variables within, Indicated in the evaluation window All power grid environmental variables within;

[0025] All integrated input sequences This constitutes the training dataset.

[0026] Step S3 describes the construction of an initial model for estimating the performance of the target energy storage power station based on a state-space, state-coupled, and neural operator scheme. This specifically includes the following steps:

[0027] The initial model for estimating the performance of the target energy storage power station includes a state space coding module, a state coupling module, and a performance estimation module connected in series.

[0028] Based on the state space, a state space coding module is constructed; the state space coding module is used to calculate the state output representation of the target energy storage power station.

[0029] Based on the state coupling scheme, a state coupling module is constructed. The state coupling module is used to perform state coupling on the state output representation of the target energy storage power station and calculate the comprehensive dynamic state representation of the target energy storage power station.

[0030] Based on the neural operator scheme, a performance estimation module is constructed. The performance estimation module is used to calculate the energy storage response state estimation result of the target energy storage power station based on the comprehensive dynamic state characterization of the target energy storage power station.

[0031] The processing procedure of the state-space coding module includes the following steps:

[0032] Define the comprehensive hidden state of the target energy storage power station at time t. for ,in This is a power fast response hidden state. This represents the evolution of a charged state into a hidden state. This is a hidden state due to temperature accumulation. This is a latent state of health decline;

[0033] For each type of hidden state, a selective state-space update equation is constructed, expressed as:

[0034] In the formula Let be the hidden state of the j-th type of state at time t. The value is , , and ; These are state transition parameters. , Let j be the selective time step corresponding to the j-th state. , For the softplus function, This is the first weight matrix. For the first bias term, The state transition matrix is ​​in continuous form. , To construct a diagonal matrix, This is the second weight matrix. This is the second bias term; For input mapping parameters, , for An identity matrix of order 1. The input mapping matrix in continuous form. For matrix dimension reconstruction operations, For the input mapping weight matrix, For input mapping bias terms, Let be the dimension of the j-th class of hidden states. Let the j-th type of input variable dimension be ; To output the mapping parameters, , To output the mapping weight matrix, To output the mapping bias term, Let the output representation dimension be the j-th class; This represents the output of the j-th state at time t. Let j be the input variable for the j-th state. hour , hour , For rated capacity, For charging efficiency, For discharge efficiency, hour , For equivalent heat loss, hour , Depth of discharge, This represents the equivalent number of iterations.

[0035] The final calculation yields the power fast response state output characterization. State-state evolution and output characterization Temperature accumulation state output characterization and health degradation state output characterization .

[0036] The processing procedure of the state coupling module includes the following steps:

[0037] The comprehensive dynamic state characterization of the target energy storage power station is calculated using the following formula. :

[0038] In the formula The power fast response state weight matrix to be learned; Here is the state weight matrix for the evolution of the charged state to be learned; The temperature cumulative state weight matrix to be learned; The weight matrix for the healthy degenerate states to be learned; The weight matrix represents the comprehensive dynamic state representation to be learned; This is the element-wise multiplication operation for Hadamard.

[0039] The performance estimation module's processing steps include the following:

[0040] Setting the energy storage response state estimation results for ;in, To estimate the active power at time t, To estimate the reactive power at time t, To estimate the state of charge at time t, To estimate the battery temperature at time t, To estimate the health status at time t, To estimate the equivalent internal resistance at time t;

[0041] The energy storage response state estimation result of the target energy storage power station is calculated using the following formula. :

[0042] In the formula The output layer mapping weight matrix; The non-linear activation function is set (ReLU, Tanh or Sigmoid functions can be used); This is the hidden layer mapping weight matrix; This refers to the hidden layer bias term; This is the output layer bias term.

[0043] The training described in step S4 specifically includes the following steps:

[0044] The following formula is used to represent the state-of-charge evolution of the target energy storage power station:

[0045] In the formula Let be the charging power at time t; Let be the discharge power at time t; The sampling interval;

[0046] The uniformity residual of the state of charge was calculated. for ;

[0047] Constructing state-of-charge constraints for ;in, It is the square of the L2 norm;

[0048] The temperature change of the energy storage battery satisfies the following expression:

[0049] In the formula T represents the equivalent heat capacity; T is the temperature of the energy storage battery. Equivalent thermal resistance;

[0050] Discretize to obtain ;

[0051] Calculated temperature uniformity residuals for ;

[0052] Constructing temperature-constrained loss for ;

[0053] The health state evolution of the target energy storage power station is represented by the following formula:

[0054] In the formula The first degradation factor is set; This is the second degradation coefficient that is set; This refers to the charge / discharge rate; Activation energy; It is the gas constant;

[0055] The health status consistency residuals were calculated. for ;

[0056] Constructing health loss constraints for ;

[0057] Construct boundary constraint loss , is represented as:

[0058] In the formula The upper limit of the allowed state of charge; The lower limit value allowed for the charged platen is set; This is the set upper limit for permissible active power. The set upper limit for permissible reactive power; The set upper limit of the allowable temperature; The set maximum allowable current value; The set minimum value for the allowed health status;

[0059] Constructing the evaluation task error loss , is represented as:

[0060] In the formula This is the first weighting coefficient set. This is the normalized reference value for active power; This is the second weighting coefficient that is set; This is the normalized reference value for reactive power; This is the set third weighting coefficient; This is the fourth weighting coefficient that is set; This is the normalized baseline value for the temperature of the energy storage battery; The fifth weighting coefficient is set; The set sixth weighting coefficient; This is the normalized reference value for the equivalent internal resistance;

[0061] Constructing state continuity constraint loss for ;

[0062] The following formula is used as the loss function:

[0063] In the formula The value of the loss function; The set weight for the state of charge loss; The set temperature loss weight; Weights for the set health status loss; The boundary loss weights are set; The set continuity loss weight;

[0064] During training, the loss function described above is used to train the model.

[0065] This invention also provides a system for implementing the aforementioned energy storage power station performance estimation method, comprising a data acquisition module, a data processing module, a model building module, a model training module, and a performance estimation module; the data acquisition module, data processing module, model building module, model training module, and performance estimation module are connected in series; the data acquisition module is used to acquire historical data information of the target energy storage power station and upload the data information to the data processing module; the data processing module is used to construct energy storage state variables, scheduling action variables, and grid environment variables of the target energy storage power station based on the received data information and the acquired data information, in order to construct a training dataset, and upload the data information to the model building module; the model building module is used to construct an initial performance estimation model of the target energy storage power station based on the received data information, using a state-space, state-coupled, and neural operator scheme, and upload the data information to the model training module; the model training module is used to train the constructed model using the obtained training dataset based on the received data information, to obtain the performance estimation model of the target energy storage power station, and upload the data information to the performance estimation module; the performance estimation module is used to complete the performance estimation of the target energy storage power station using the obtained model based on the received data information.

[0066] The energy storage power station performance estimation method and system provided by this invention acquires and processes various comprehensive information of the target energy storage power station to construct a training dataset, and constructs and trains a relatively simple energy storage power station performance estimation model based on state space, state coupling and neural operator schemes. This not only enables the performance estimation of energy storage power stations, but also has higher reliability and better accuracy. Attached Figure Description

[0067] Figure 1 This is a schematic diagram of the method flow of the present invention.

[0068] Figure 2 This is a schematic diagram of the key state curves for frequency regulation operation of an energy storage power station, as shown in the embodiment of the method of the present invention.

[0069] Figure 3 This is a schematic diagram of the functional modules of the system of the present invention. Detailed Implementation

[0070] like Figure 1 The diagram shown is a flowchart of the method of the present invention: The energy storage power station performance estimation method disclosed in this invention includes the following steps:

[0071] S1. Obtain historical data information of the target energy storage power station; specifically including the following steps:

[0072] Obtain historical data information of the target energy storage power station;

[0073] The data information includes the active power data, reactive power data, state of charge data, health status data, battery temperature data, current data, and terminal voltage data of the target energy storage power station; active power command data, reactive power command data, ramp speed requirement data, and operation mode data received by the target energy storage power station from the grid it is connected to; and frequency data, grid connection point voltage data, frequency deviation data, voltage deviation data, load power data, renewable energy output data, and ambient temperature data of the grid to which the target energy storage power station is connected.

[0074] S2. Based on the data obtained in step S1, construct the energy storage state variables, scheduling action variables, and grid environment variables of the target energy storage power station to build a training dataset; specifically, this includes the following steps:

[0075] To address the challenges of fragmented input sources, inconsistent key influencing factors across different operating scenarios, and the inability of a single operating indicator to fully represent the current support capabilities of energy storage power stations during performance evaluation, this invention first constructs a multi-source input variable system based on the current evaluation time, encompassing energy storage operating status, dispatch actions, grid environment, and equipment health status. This approach integrates the energy storage power station's own state, grid-side disturbances, dispatch requirements, and equipment health constraints into a unified evaluation framework, providing a foundation for subsequent multi-timescale state modeling and comprehensive performance evaluation.

[0076] Construct the energy storage state variables of the target energy storage power station at time t :

[0077] In the formula Let be the active power of the target energy storage power station at time t; Let be the reactive power of the target energy storage power station at time t; The state of charge of the target energy storage power station at time t; The health status of the target energy storage power station at time t; Let be the battery temperature of the target energy storage power station at time t; Let be the equivalent internal resistance of the target energy storage power station at time t; Let be the current of the target energy storage power station at time t; Let be the terminal voltage of the target energy storage power station at time t;

[0078] Construct the scheduling action variables of the target energy storage power station at time t. :

[0079] In the formula The active power command received by the target energy storage power station at time t; The reactive power command received by the target energy storage power station at time t; The required ramp rate for the target energy storage power station at time t; The variable is used to identify the operating mode of the target energy storage power station at time t. The operating modes include frequency regulation mode, peak shaving mode, standby mode, voltage support mode, standby mode, and power limiting mode. This variable is used to characterize the scheduling requirements and task types of the energy storage power station within the current evaluation window.

[0080] Construct the grid environment variables of the target energy storage power station at time t :

[0081] In the formula The frequency of the power grid to which the target energy storage power station is connected at time t; The grid connection point voltage of the power grid to which the target energy storage power station is connected at time t; The frequency deviation of the power grid to which the target energy storage power station is connected at time t; The voltage deviation of the power grid to which the target energy storage power station is connected at time t; The load power of the power grid connected to the target energy storage power station at time t; The renewable energy output value of the power grid connected to the target energy storage power station at time t; The ambient temperature of the power grid to which the target energy storage power station is connected at time t; by using this power grid environmental variable, external conditions such as frequency deviation, voltage fluctuation, load change and new energy output fluctuation can be incorporated into the evaluation process, avoiding one-sided evaluation based solely on the state of the energy storage equipment itself;

[0082] Construct the evaluation window at time t for ,in The length of the evaluation window; by setting the evaluation window, the power changes, command responses, state of charge changes, temperature changes and health status changes of the energy storage power station over a continuous period of time can be included in the same evaluation process, avoiding the influence of instantaneous fluctuations on the evaluation results caused by using only single-point data;

[0083] Construct the comprehensive input sequence of the target energy storage power station at time t for ,in Indicated in the evaluation window All energy storage state variables within, Indicated in the evaluation window All scheduling action variables within, Indicated in the evaluation window All grid environment variables within the current evaluation window; this input sequence is used to characterize the operating status, dispatching role, grid environment and equipment health level of the energy storage power station within the current evaluation window; thus, it can avoid evaluating based on a single indicator or data at a single moment, and enable subsequent models to simultaneously consider the impact of the operating process, dispatching requirements, grid disturbances and health constraints on the performance of the energy storage power station.

[0084] All integrated input sequences This constitutes the training dataset;

[0085] S3. Based on the state-space, state-coupled, and neural operator schemes, construct an initial model for performance estimation of the target energy storage power station; specifically including the following steps:

[0086] The initial model for estimating the performance of the target energy storage power station includes a state space coding module, a state coupling module, and a performance estimation module connected in series.

[0087] Based on the state space, a state space coding module is constructed; the state space coding module is used to calculate the state output representation of the target energy storage power station.

[0088] Based on the state coupling scheme, a state coupling module is constructed. The state coupling module is used to perform state coupling on the state output representation of the target energy storage power station and calculate the comprehensive dynamic state representation of the target energy storage power station.

[0089] Based on the neural operator scheme, a performance estimation module is constructed. The performance estimation module is used to calculate the energy storage response state estimation result of the target energy storage power station based on the comprehensive dynamic state characterization of the target energy storage power station.

[0090] In practice, the state space coding module's processing steps include the following:

[0091] Define the comprehensive hidden state of the target energy storage power station at time t. for ,in This is a power fast response hidden state. This represents the evolution of a charged state into a hidden state. This is a hidden state due to temperature accumulation. This is a latent state of health decline;

[0092] For each type of hidden state, a selective state-space update equation is constructed, expressed as:

[0093] In the formula Let be the hidden state of the j-th type of state at time t. The value is , , and ; These are state transition parameters. , Let j be the selective time step corresponding to the j-th state. , For the softplus function, This is the first weight matrix. For the first bias term, The state transition matrix is ​​in continuous form. , To construct a diagonal matrix, This is the second weight matrix. This is the second bias term; For input mapping parameters, , for An identity matrix of order 1. The input mapping matrix in continuous form. For matrix dimension reconstruction operations, For the input mapping weight matrix, For input mapping bias terms, Let be the dimension of the j-th class of hidden states. Let the j-th type of input variable dimension be ; To output the mapping parameters, , To output the mapping weight matrix, To output the mapping bias term, Let the output representation dimension be the j-th class; This represents the output of the j-th state at time t. Let j be the input variable for the j-th state. hour , hour , For rated capacity, For charging efficiency, For discharge efficiency, hour , For equivalent heat loss, hour , Depth of discharge, This represents the equivalent number of iterations.

[0094] Used to characterize the power tracking, rapid adjustment and support response capabilities of energy storage power stations under dispatch commands and grid disturbances; Used to characterize the changes in the state of charge and the current available energy level of an energy storage power station during the charging and discharging process; Used to characterize the effects of temperature, internal resistance, depth of discharge, and cycle number on health degradation;

[0095] The final calculation yields the power fast response state output characterization. State-state evolution and output characterization Temperature accumulation state output characterization and health degradation state output characterization ;

[0096] This module does not treat all input variables as the same type of feature for unified processing. Instead, it models power response, state of charge evolution, temperature accumulation, and health degradation separately according to the operating mechanism of energy storage power stations. This allows the model to capture rapid power changes, energy state changes, hysteretic changes in thermal state, and slow health degradation processes within the current evaluation window. This approach improves the accuracy of dynamic state characterization of energy storage power stations and provides a foundation for subsequent comprehensive performance evaluation, physical consistency correction, and anomaly cause identification.

[0097] Meanwhile, unlike traditional self-attention mechanisms, this module does not uniformly weight all features through a global attention weight matrix. Instead, it establishes state-space update processes based on the physical time scale of the energy storage power station's state variables, enabling rapid power response, state of charge evolution, temperature accumulation, and health degradation to each have independent but coupled dynamic memory structures. As a result, the module can enhance power response state updates in scenarios with rapid power changes such as frequency regulation, enhance state of charge and temperature accumulation state memories in scenarios with long-term energy regulation such as peak shaving, and enhance temperature and health state constraints in high-temperature or health degradation-sensitive scenarios, thereby improving the adaptability and interpretability of performance evaluation results under different operating scenarios.

[0098] In practical implementation, the state coupling module's processing procedure includes the following steps:

[0099] The comprehensive dynamic state characterization of the target energy storage power station is calculated using the following formula. :

[0100] In the formula The power fast response state weight matrix to be learned; Here is the state weight matrix for the evolution of the charged state to be learned; The temperature cumulative state weight matrix to be learned; The weight matrix for the healthy degenerate states to be learned; The weight matrix represents the comprehensive dynamic state representation to be learned; This is the Hadamard element-wise multiplication operation;

[0101] This module can unify and integrate state outputs at different time scales into a comprehensive dynamic state characterization, thereby depicting the impact of power fluctuations on the state of charge, the impact of power loss on temperature rise, and the impact of temperature changes and internal resistance changes on health degradation, providing a unified state input for subsequent comprehensive performance evaluation and physical consistency correction.

[0102] Through the fusion method of this module, the present invention not only retains the independent characterization of each time scale state, but also can characterize the coupling relationship such as the change of state of charge caused by high-frequency power fluctuations, the temperature rise caused by power loss, and the accelerated health degradation caused by temperature rise, thereby avoiding making a one-sided judgment on the performance of the energy storage power station based on only a single state channel.

[0103] In practice, the performance estimation module's processing steps include the following:

[0104] Setting the energy storage response state estimation results for ;in, To estimate the active power at time t, To estimate the reactive power at time t, To estimate the state of charge at time t, To estimate the battery temperature at time t, To estimate the health status at time t, To estimate the equivalent internal resistance at time t;

[0105] The energy storage response state estimation result of the target energy storage power station is calculated using the following formula. :

[0106] In the formula The output layer mapping weight matrix; The non-linear activation function is set (ReLU, Tanh or Sigmoid functions can be used); This is the hidden layer mapping weight matrix; This refers to the hidden layer bias term; For output layer bias terms;

[0107] This module is used to learn the mapping relationship between scheduling actions, power grid environment and current dynamic state, obtain the energy storage response state estimation results under the current evaluation window, and provide a basis for subsequent calculation of state of charge constraints, thermal state constraints, health degradation constraints and boundary constraints;

[0108] S4. Using the training dataset obtained in step S2, train the model constructed in step S3 to obtain the performance estimation model for the target energy storage power station; specifically including the following steps:

[0109] The present invention constructs physical consistency constraints including state of charge constraints, thermal state constraints, health degradation constraints, and operating boundary constraints, which are used to constrain and correct the model output results, so that the evaluation results are not only consistent with the evaluation labels or historical experience, but also meet the actual operating mechanism and safety boundary requirements of the energy storage power station.

[0110] The following formula is used to represent the state-of-charge evolution of the target energy storage power station:

[0111] In the formula Let be the charging power at time t; Let be the discharge power at time t; The sampling interval;

[0112] The uniformity residual of the state of charge was calculated. for ;

[0113] Constructing state-of-charge constraints for ;in, It is the square of the L2 norm;

[0114] The temperature change of the energy storage battery satisfies the following expression:

[0115] In the formula T represents the equivalent heat capacity; T is the temperature of the energy storage battery. Equivalent thermal resistance;

[0116] Discretize to obtain ;

[0117] Calculated temperature uniformity residuals for ;

[0118] Constructing temperature-constrained loss for ;

[0119] The health state evolution of the target energy storage power station is represented by the following formula:

[0120] In the formula The first degradation factor is set; This is the second degradation coefficient that is set; This refers to the charge / discharge rate; Activation energy; It is the gas constant;

[0121] The health status consistency residuals were calculated. for ;

[0122] Constructing health loss constraints for ;

[0123] Construct boundary constraint loss , is represented as:

[0124] In the formula The upper limit of the allowed state of charge; The lower limit value allowed for the charged platen is set; This is the set upper limit for permissible active power. The set upper limit for permissible reactive power; The set upper limit of the allowable temperature; The set maximum allowable current value; The set minimum value for the allowed health status;

[0125] Constructing the evaluation task error loss , is represented as:

[0126] In the formula This is the first weighting coefficient set. This is the normalized reference value for active power; This is the second weighting coefficient that is set; This is the normalized reference value for reactive power; This is the set third weighting coefficient; This is the fourth weighting coefficient that is set; This is the normalized baseline value for the temperature of the energy storage battery; The fifth weighting coefficient is set; The set sixth weighting coefficient; This is the normalized reference value for the equivalent internal resistance;

[0127] Constructing state continuity constraint loss for ;

[0128] The following formula is used as the loss function:

[0129] In the formula The value of the loss function; The set weight for the state of charge loss; The set temperature loss weight; Weights for the set health status loss; The boundary loss weights are set; The set continuity loss weight;

[0130] During training, the loss function described above is used to train the model;

[0131] Through the above training process, the model can reduce the error of the evaluation task while satisfying physical constraints such as state of charge, temperature, health degradation and operating boundary, and avoid abrupt changes in the evaluation results between adjacent evaluation windows that do not conform to the actual operating rules.

[0132] S5. Using the model obtained in step S4, complete the performance estimation of the target energy storage power station.

[0133] The effects of the method of the present invention will be illustrated below with reference to an embodiment:

[0134] This example uses a grid-connected electrochemical energy storage power station as the subject. The station has a rated power of 100MW and a rated capacity of 200MWh, primarily participating in grid frequency regulation, peak shaving reserve, and new energy consumption support. A continuous evaluation period from 10:00:00 to 10:15:00 is selected for analysis, with a data sampling period of 1 second, resulting in 900 sampling points. This example uses actual operating data and equipment boundary parameters within the current evaluation window as input and does not involve future task prediction or future operating trajectory generation.

[0135] The input data for the simulation includes dispatch commands, energy storage operating status, grid environment, and equipment boundary parameters. Specifically, the dispatch active power command range is -60MW to 70MW, with a command change rate requirement of 2.5MW / s; the actual active power range of the energy storage is -58MW to 66MW, the reactive power range is -10Mvar to 12Mvar, the state of charge range is 24.8% to 31.2%, the health status is 91.6%, and the average battery temperature is 31.5°C. up to 39.8 The highest temperature was 42.3 degrees Celsius. The grid connection frequency range is 49.94Hz to 50.06Hz, the frequency deviation range is -0.06Hz to 0.06Hz, and the ambient temperature is 28.5°C. The equipment boundary conditions are set as follows: minimum charge level 20%, maximum charge level 90%, and temperature warning threshold 40°C. Temperature limit 45 The health condition is at least 80%, and the rated capacity is 200MWh.

[0136] Figure 2 Key state curves for the frequency regulation operation of the energy storage power station during the evaluation period are presented, including curves showing changes in power command and actual active power, changes in state of charge, and changes in battery temperature. Figure 2 It can be seen that the dispatch power command fluctuated within the range of approximately -60MW to 70MW. The actual active power generally tracked the command change trend well, but there was some lag and deviation during periods of rapid change, indicating that the energy storage power station possesses good frequency regulation response and ramp-up capabilities. Simultaneously, the state of charge (SOC) gradually decreased from approximately 31.2% to around 24.8%, indicating that the energy storage power station was mainly in a discharge support state during this period, and the SOC was gradually approaching its lower limit; the average battery temperature decreased from approximately 31.5℃. It rose to 39.8 The highest temperature rose to 42.3 degrees Celsius. It is approaching 40 The warning threshold has not yet been reached (45). The upper temperature limit indicates that there is a certain risk of temperature accumulation during continuous frequency modulation support.

[0137] Under the aforementioned real-time conditions, the operating status of the energy storage power station during the current evaluation period was assessed using the method of this invention. The results show that the energy storage power station's frequency regulation response capability score is 88.2, and its ramp-up capability score is 84.0, indicating good command tracking and rapid response capabilities. However, due to the state of charge being close to the lower limit and the temperature approaching the warning zone, its continuous support capability score is only 63.5, resulting in a comprehensive performance score of 78.6 and an overall rating of medium. Further calculations show that the currently recommended maximum available power is 60MW, and the available electricity in the current discharge direction is 9.6MWh. Under 60MW discharge support, it can sustain operation for approximately 9.6 minutes, and under 40MW discharge support, it can sustain operation for approximately 14.4 minutes.

[0138] Therefore, the present invention can provide a comprehensive performance evaluation while further identifying the current available capacity and main limiting factors of the energy storage power station. Compared with judging solely based on power point tracking performance, this method can simultaneously reflect power response capability, state of charge boundary, temperature safety margin, and continuous support capability, avoiding overestimation of the current actual available capacity of the energy storage power station, and providing a basis for dispatching power-limited calls, state of charge recovery, and temperature risk management.

[0139] like Figure 3The diagram shows the functional modules of the system of the present invention: The system for implementing the energy storage power station performance estimation method disclosed in this invention includes a data acquisition module, a data processing module, a model building module, a model training module, and a performance estimation module; the data acquisition module, data processing module, model building module, model training module, and performance estimation module are connected in series; the data acquisition module is used to acquire historical data information of the target energy storage power station and upload the data information to the data processing module; the data processing module is used to construct energy storage state variables, scheduling action variables, and grid environment variables of the target energy storage power station based on the received data information and the acquired data information, in order to construct a training dataset, and upload the data information to the model building module; the model building module is used to construct an initial performance estimation model of the target energy storage power station based on the received data information, using a state space, state coupling, and neural operator scheme, and upload the data information to the model training module; the model training module is used to train the constructed model using the obtained training dataset based on the received data information, to obtain the performance estimation model of the target energy storage power station, and upload the data information to the performance estimation module; the performance estimation module is used to complete the performance estimation of the target energy storage power station using the obtained model based on the received data information.

Claims

1. A method for estimating the performance of an energy storage power station, comprising the following steps: S1. Obtain historical data information of the target energy storage power station; S2. Based on the data obtained in step S1, construct the energy storage state variables, scheduling action variables, and power grid environment variables of the target energy storage power station to build a training dataset; S3. Based on the state-space, state-coupled, and neural operator schemes, construct an initial model for performance estimation of the target energy storage power station; S4. Using the training dataset obtained in step S2, train the model constructed in step S3 to obtain the performance estimation model of the target energy storage power station; S5. Using the model obtained in step S4, complete the performance estimation of the target energy storage power station.

2. The energy storage power station performance estimation method according to claim 1, characterized in that... Step S1, which involves obtaining historical data information of the target energy storage power station, specifically includes the following steps: Obtain historical data information of the target energy storage power station; The data information includes the active power data, reactive power data, state of charge data, health status data, battery temperature data, current data, and terminal voltage data of the target energy storage power station; active power command data, reactive power command data, ramp speed requirement data, and operating mode data received by the target energy storage power station from the connected power grid; and frequency data, grid connection point voltage data, frequency deviation data, voltage deviation data, load power data, renewable energy output data, and ambient temperature data of the power grid to which the target energy storage power station is connected.

3. The energy storage power station performance estimation method according to claim 2, characterized in that... Step S2, which involves constructing the energy storage state variables, scheduling action variables, and grid environment variables of the target energy storage power station based on the data information obtained in step S1, to build a training dataset, specifically includes the following steps: Construct the energy storage state variables of the target energy storage power station at time t : In the formula Let be the active power of the target energy storage power station at time t; Let be the reactive power of the target energy storage power station at time t; The state of charge of the target energy storage power station at time t; The health status of the target energy storage power station at time t; Let be the battery temperature of the target energy storage power station at time t; Let be the equivalent internal resistance of the target energy storage power station at time t; Let be the current of the target energy storage power station at time t; Let be the terminal voltage of the target energy storage power station at time t; Construct the scheduling action variables of the target energy storage power station at time t. : In the formula The active power command received by the target energy storage power station at time t; The reactive power command received by the target energy storage power station at time t; The required ramp rate for the target energy storage power station at time t; The variable is the operating mode identifier of the target energy storage power station at time t. The operating modes include frequency regulation mode, peak shaving mode, standby mode, voltage support mode, standby mode, and power limiting operation mode. Construct the grid environment variables of the target energy storage power station at time t : In the formula The frequency of the power grid to which the target energy storage power station is connected at time t; The grid connection point voltage of the power grid to which the target energy storage power station is connected at time t; The frequency deviation of the power grid to which the target energy storage power station is connected at time t; The voltage deviation of the power grid to which the target energy storage power station is connected at time t; The load power of the power grid connected to the target energy storage power station at time t; The renewable energy output value of the power grid connected to the target energy storage power station at time t; The ambient temperature of the power grid to which the target energy storage power station is connected at time t; Construct the evaluation window at time t for ,in The length of the evaluation window; Construct the comprehensive input sequence of the target energy storage power station at time t for ,in Indicated in the evaluation window All energy storage state variables within, Indicated in the evaluation window All scheduling action variables within, Indicated in the evaluation window All power grid environmental variables within; All integrated input sequences This constitutes the training dataset.

4. The energy storage power station performance estimation method according to claim 3, characterized in that... Step S3 describes the construction of an initial model for estimating the performance of the target energy storage power station based on a state-space, state-coupled, and neural operator scheme. This specifically includes the following steps: The initial model for estimating the performance of the target energy storage power station includes a state space coding module, a state coupling module, and a performance estimation module connected in series. Based on the state space, a state space coding module is constructed; the state space coding module is used to calculate the state output representation of the target energy storage power station. Based on the state coupling scheme, a state coupling module is constructed; The state coupling module is used to perform state coupling on the state output representation of the target energy storage power station and calculate the comprehensive dynamic state representation of the target energy storage power station. A performance estimation module is constructed based on a neural operator scheme. The performance estimation module is used to calculate the energy storage response state estimation result of the target energy storage power station based on the comprehensive dynamic state characterization of the target energy storage power station.

5. The energy storage power station performance estimation method according to claim 4, characterized in that... The processing procedure of the state-space coding module includes the following steps: Define the comprehensive hidden state of the target energy storage power station at time t. for ,in This is a power fast response hidden state. This represents the evolution of a charged state into a hidden state. This is a hidden state due to temperature accumulation. This is a latent state of health decline; For each type of hidden state, a selective state-space update equation is constructed, expressed as: In the formula Let be the hidden state of the j-th type of state at time t. The value is , , and ; These are state transition parameters. , Let j be the selective time step corresponding to the j-th state. , For the softplus function, This is the first weight matrix. For the first bias term, The state transition matrix is ​​in continuous form. , To construct a diagonal matrix, This is the second weight matrix. This is the second bias term; For input mapping parameters, , for An identity matrix of order 1. The input mapping matrix in continuous form. For matrix dimension reconstruction operations, For the input mapping weight matrix, For input mapping bias terms, Let be the dimension of the j-th class of hidden states. Let the j-th type of input variable dimension be ; To output the mapping parameters, , To output the mapping weight matrix, To output the mapping bias term, Let the output representation dimension be the j-th class; This represents the output of the j-th state at time t. Let j be the input variable for the j-th state. hour , hour , For rated capacity, For charging efficiency, For discharge efficiency, hour , For equivalent heat loss, hour , Depth of discharge, This represents the equivalent number of iterations. The final calculation yields the power fast response state output characterization. State-state evolution and output characterization Temperature accumulation state output characterization and health degradation state output characterization .

6. The energy storage power station performance estimation method according to claim 5, characterized in that... The processing procedure of the state coupling module includes the following steps: The comprehensive dynamic state characterization of the target energy storage power station is calculated using the following formula. : In the formula The power fast response state weight matrix to be learned; Here is the state weight matrix for the evolution of the charged state to be learned; The temperature cumulative state weight matrix to be learned; The weight matrix for the healthy degenerate states to be learned; The weight matrix represents the comprehensive dynamic state representation to be learned; This is the element-wise multiplication operation for Hadamard.

7. The energy storage power station performance estimation method according to claim 6, characterized in that... The performance estimation module's processing steps include the following: Setting the energy storage response state estimation results for ;in, To estimate the active power at time t, To estimate the reactive power at time t, To estimate the state of charge at time t, To estimate the battery temperature at time t, To estimate the health status at time t, To estimate the equivalent internal resistance at time t; The energy storage response state estimation result of the target energy storage power station is calculated using the following formula. : In the formula The output layer mapping weight matrix; The nonlinear activation function is defined. This is the hidden layer mapping weight matrix; This refers to the hidden layer bias term; This is the output layer bias term.

8. The energy storage power station performance estimation method according to claim 7, characterized in that... The training described in step S4 specifically includes the following steps: The following formula is used to represent the state-of-charge evolution of the target energy storage power station: In the formula Let be the charging power at time t; Let be the discharge power at time t; The sampling interval; The uniformity residual of the state of charge was calculated. for ; Constructing state-of-charge constraints for ;in, It is the square of the L2 norm; The temperature change of the energy storage battery satisfies the following expression: In the formula T represents the equivalent heat capacity; T is the temperature of the energy storage battery. Equivalent thermal resistance; Discretize to obtain ; Calculated temperature uniformity residuals for ; Constructing temperature-constrained loss for ; The health state evolution of the target energy storage power station is represented by the following formula: In the formula The first degradation factor is set; This is the second degradation coefficient that is set; This refers to the charge / discharge rate; Activation energy; It is the gas constant; The health status consistency residuals were calculated. for ; Constructing health loss constraints for ; Construct boundary constraint loss , is represented as: In the formula The upper limit of the allowed state of charge; The lower limit value allowed for the charged platen is set; This is the set upper limit for permissible active power. The set upper limit for permissible reactive power; The set upper limit of the allowable temperature; The set maximum allowable current value; The set minimum value for the allowed health status; Constructing the evaluation task error loss , is represented as: In the formula This is the first weighting coefficient set. This is the normalized reference value for active power; This is the second weighting coefficient that is set; This is the normalized reference value for reactive power; This is the set third weighting coefficient; This is the fourth weighting coefficient that is set; This is the normalized baseline value for the temperature of the energy storage battery; The fifth weighting coefficient is set; The set sixth weighting coefficient; This is the normalized reference value for the equivalent internal resistance; Constructing state continuity constraint loss for ; The following formula is used as the loss function: In the formula The value of the loss function; The set weight for the state of charge loss; The set temperature loss weight; Weights for the set health status loss; The boundary loss weights are set; The set continuity loss weight; During training, the loss function described above is used to train the model.

9. A system for implementing the energy storage power station performance estimation method according to any one of claims 1 to 8, characterized in that... It includes a data acquisition module, a data processing module, a model building module, a model training module, and a performance estimation module; these modules are connected in series. The data acquisition module is used to acquire historical data information of the target energy storage power station and upload the data information to the data processing module. The data processing module is used to construct energy storage state variables, scheduling action variables, and grid environment variables of the target energy storage power station based on the received data information and the acquired data information, in order to build a training dataset, and then upload the data information to the model building module. The model building module is used to construct an initial performance estimation model for the target energy storage power station based on the received data information, state space, state coupling and neural operator schemes, and upload the data information to the model training module; The model training module is used to train the constructed model based on the received data information and the obtained training dataset to obtain the performance estimation model of the target energy storage power station, and upload the data information to the performance estimation module. The performance estimation module is used to perform performance estimation of the target energy storage power station based on the received data and the obtained model.

Images